The project has addressed the following areas in the past year: 1. Association of highly charged intrinsically disordered proteins. Following up on work done last year characterizing the equilibrium association of the complex between histone H1 and the protein prothymosin alpha, we have now addressed the kinetics of association and dissociation (collaboration with Ben Schuler (University of Zurich) and Birthe Kragelund (University of Copenhagen)). The prothymosin appears to be exchanged by an associative mechanism whereby association of a second prothymosin molecule displaces the first. This finding explains the relatively rapid dissociation kinetics (microsecond time scale) compare to what would be expected based on the diffusion-limited association and picomolar binding. Coarse grained models have been used to describe the exchange mechanism and to calculate a free energy surface for binding. This work is currently being prepared for publication. (R. Best) 2. Role of prothymosin as a chaperone for histone H1 on the nucleosome. Related to (1) above, we have been investigating the ability of prothymosin to facilitate dissociation of histone H1 from the nucleosome. Coarse-grained models have been used to describe the bound ensemble of H1 to the nucleosome, reproducing the experimental FRET efficiencies. Using a novel methodology, we have also computed the effect of prothymosin on the H1 dissociation rates, finding an acceleration of around 2 orders of magnitude, in excellent agreement with experimental estimates (R. Best, D. Mercadante). Publication in preparation. (R. Best) 3. Role of viral capsid protein as nucleic acid folding chaperone (collaboration with B. Schuler). We have developed a very basic coarse-grained model for nucleic acids and protein-nucleic acid interactions in order to examine the effect of a disordered protein on the folding of a model DNA hairpin. The simulations showed that the electrostatic screening effect of the protein, resulting in a collapse of the nucleic acid was sufficient to explain all of the acceleration in observed folding rate (R. Best). (1) 4. All-atom and coarse-grained force field development. 4.1. Improving hydrogen bond potential in all-atom models of RNA folding (collaboration with P. Kuhrova). Our earlier work had shown that the hydrogen bond potential was a possible weak point in all-atom protein force fields and may be responsible for their inability to distinguish folded from misfolded structures. We have now developed a new hydrogen bond potential with improved performanc for RNA folding. (2) 4.2. Local vs global effects on all-atom protein force fields. We have recently shown that certain experimental observables are almost exclusively sensitive to local structural properties of proteins, while others capture more global properties. Both types of observables therefore need to be considered when assessing protein force fields (Collaboration with J. Mittal) (3). 4.3. Improving the coarse-grained Martini force field (collaboration with J. Domanski, P. Telles de Souza). We had earlier shown that the current version of the Martini coarse-grained model could not capture the native state of glycophorin (a prototype for transmembrane helix association) as the most stable bound state. In collaboration with the Martini developers we are testing the next generation of Martini force field, which shows a much improved reproduction of the native structure. (R. Best) 5. Multidomain protein misfolding. Following on earlier work in which we investigated multidomain protein misfolding using coarse-grained models, we have now developed an easy to use tool (TADOSS) to predict misfolding propensity based on a phenomenological model in that work (Collaboration with A. Bateman) (4,5). We are also currently collaborating with Alex Bateman (Cambridge) and Jennifer Potts (York) on how certain multidomain bacterial proteins are able to avoid misfolding despite the adjacent domains having high sequence similarity. (P. Tian, R. Best) 6. Co-translational protein folding. Last year, we developed a coarse-grained model of the ribosome suitable for studying co-translational folding, and applied it to the folding of the I27 domain of titin (6), the src SH3 domain (7) and several other proteins (8). This year, in collaboration with Gunnar von Heijne, we used the model to investigate more directly the relationship between the forces arising from the folding nascent chain and the yield of full length protein obtained in arrest peptide experiments. We also devised a method for obtaining these forces directly from experiment by using a series of different arrest peptides with the same protein constructs (Manuscript under revision in PNAS). We are also collaborating with Susan Marqusee on the co-translational folding of RNase H, and the effect of mutants which make the folding either more or less cooperative. In another collaboration with the group of Sander Tans, we are using our coarse-grained co-translational folding model to understand the effects of the ribosome on the folding and unfolding rates of ADR1a in the ribosome exit tunnel, as probed by single molecule force spectrosccopy (R. Best, P. Tian). 7. Environmental effects on phase separation of intrinsically disordered proteins. We are working to further develop our coarse-grained model for liquid-liquid phase separation (LLPS) of intrinsically disordered proteins published last year, to include such effects as the addition of ions (Hofmeister effects) and co-solvents, which are known to affect the phase diagram for LLPS in vitro. (T. Dannenhofer-Lafage) 8. Modelling sensitivity of single molecule experiments to protein folding transition paths using molecular simulations. Recent single molecule fluorescence experiments have been able to detect transition paths between folded and unfolded states of proteins by combining photon by photon detection with sophisticated maximum likelihood analysis algorithms. However, it is not clear how the inferred transition path durations relate to the actual folding transition path lengths, since they cannot be independently measured. We have used simulations as a model to generate coarse-grained folding trajectories for two proteins (alpha3D, protein G), in which we can unambiguously assign transition paths. We then generated synthetic photon trajectories from these simulations and analyzed them in the same way as the experimental data. We found that the experimentally inferred transition path durations are of the right magnitude, but systematically shorter than the true durations (G. Taumoefolau). Group members involved in each project are listed at the end of each section.